Unable to complete install with CUDA (make install_cuda) about gocv HOT 3 CLOSED

Bug: Unknown argument -j error when building OpenCV with CUDA using make install_cuda

Description:
I encountered an issue while trying to install GoCV with CUDA support by running sudo make install_cuda. During the build process, the command fails with the following error message:

CMake Error: Unknown argument -j
CMake Error: Run 'cmake --help' for all supported options.
make[1]: *** No targets specified and no makefile found.  Stop.
...
make: *** No rule to make target 'install'.  Stop.

After inspecting the Makefile, I observed that the cmake command within the build_cuda target incorrectly uses the -j argument, which is not a valid option for cmake.

Here is the relevant excerpt from the Makefile:

build_cuda:
	cd $(TMP_DIR)opencv/opencv-$(OPENCV_VERSION)
	mkdir build
	cd build
	rm -rf *
	cmake -j $(shell nproc --all --ignore 1) -D CMAKE_BUILD_TYPE=RELEASE ...
	$(MAKE) -j $(shell nproc --all --ignore 1)
	$(MAKE) preinstall
	cd -

Proposed Fix:
The -j $(shell nproc --all --ignore 1) should be removed from the cmake command. The correct way to handle parallel builds is to use the -j option with make (which is already done correctly in the Makefile).

Here is the corrected version of the build_cuda target:

build_cuda:
	cd $(TMP_DIR)opencv/opencv-$(OPENCV_VERSION)
	mkdir build
	cd build
	rm -rf *
	cmake -D CMAKE_BUILD_TYPE=RELEASE \
		-D CMAKE_INSTALL_PREFIX=/usr/local \
		-D BUILD_SHARED_LIBS=${BUILD_SHARED_LIBS} \
		-D OPENCV_EXTRA_MODULES_PATH=$(TMP_DIR)opencv/opencv_contrib-$(OPENCV_VERSION)/modules \
		-D BUILD_DOCS=OFF \
		-D BUILD_EXAMPLES=OFF \
		-D BUILD_TESTS=OFF \
		-D BUILD_PERF_TESTS=ON \
		-D BUILD_opencv_java=NO \
		-D BUILD_opencv_python=NO \
		-D BUILD_opencv_python2=NO \
		-D BUILD_opencv_python3=NO \
		-D WITH_JASPER=OFF \
		-D WITH_TBB=ON \
		-DOPENCV_GENERATE_PKGCONFIG=ON \
		-DWITH_CUDA=ON \
		-DENABLE_FAST_MATH=1 \
		-DCUDA_FAST_MATH=1 \
		-DWITH_CUBLAS=1 \
		-DCUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda/ \
		-DBUILD_opencv_cudacodec=OFF \
		-D WITH_CUDNN=ON \
		-D OPENCV_DNN_CUDA=ON \
		-D CUDA_ARCH_BIN=7.5 \
		-D CUDA_GENERATION=Auto ..
	$(MAKE) -j $(shell nproc --all --ignore 1)
	$(MAKE) preinstall
	cd -

from gocv.

So, I has two issues here, the first one was that the -j $(shell nproc --all --ignore 1) should be removed from the cmake command in the build_cuda step.

the second one, was that I had my cuda directory at /usr/local/cuda-12/, and the Makefile is pointing to /usr/local/cuda/. So I just changed the route and it compiled successfully

from gocv.

just checked the DNN Pose Detection example setting CUDA as default backend, it works way better now comparing to the CPU, so I'm going to close the issue myself as it is finally solved 🚀

from gocv.